Online reinforcement learning with sparse rewards through an active inference capsule.
Alejandro Daniel NoelCharel van HoofBeren MillidgePublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- markov decision processes
- online learning
- state space
- function approximation
- reward shaping
- optimal policy
- model free
- reinforcement learning algorithms
- high dimensional
- reward function
- optimal control
- balancing exploration and exploitation
- inference process
- sparse data
- real time
- probabilistic inference
- bayesian networks
- learning algorithm
- learning classifier systems
- action selection
- temporal difference
- multi agent
- markov decision process
- average reward
- batch mode
- machine learning